1. Field of the Invention
The invention is related to the field of communications, and in particular, to communication systems that consider the cost of back-up links in a protection scheme.
2. Description of the Prior Art
FIG. 1 illustrates communication system 100 in an example of the prior art. Communication system 100 includes nodes 105-106 and paths 107-108. Node 105 includes routers 101-102 that form a first mated pair. Node 106 includes routers 103-104 that form a second mated pair. Path 107 includes links 111-112. Path 108 includes links 113-114. Link 111 couples router 101 to router 103. Link 112 couples router 102 to router 104. Link 113 couples router 101 to router 103. Link 114 couples router 102 to router 104.
Paths 107-108 are geographically diverse to provide path diversity if one of the paths fails. In this example, path 107 is geographically shorter than path 108—possibly by thousands of miles. In some cases, nodes 105 and 106 are on a communication ring where short path 107 represents a short segment of the ring between nodes 105-106, and long path 108 represents the longer segment around the other side of the ring.
In a normal operating mode, links 111-114 are each loaded to 40% of capacity. Thus, half of the traffic between nodes 105-106 traverses long path 108 in the normal operating mode. If router 101 fails, then mated router 102 takes over for router 101, so that router failure is handled at layer 2/3 of the Open Systems Interconnection (OSI) stack. In this failure mode, the load of links 111 and 113 drops to zero since these links are coupled to failed router 101, and correspondingly, the load of links 112 and 114 rises from 40% to 80%, because these links now carry the added load from unused links 111 and 113. Note that half of the traffic still takes the long path 108.
If link 111 fails, then router 101 transfers the traffic over link 113, so that link failure is also handled at OSI layer 2/3. In this failure mode, the load of failed link 111 drops to zero, and correspondingly, the load of link 113 rises from 40% to 80% since link 113 now carries the added load of failed link 111.
Router 101 has a carrier delay timer that starts after a loss of signal is detected, such as OSI layer 1 detection. The carrier delay timer must time out before the above-described OSI layer 2/3 restoration is implemented. The carrier delay timer prevents layer 2/3 restoration from occurring in response to a mere signal glitch where a quality signal quickly returns. The timer is set relatively low, such as 20 milliseconds.
The expense of links 111-114 can be measured by a fixed cost per mile, and thus, long links are more expensive than short links. Links 113-114 follow long path 108, which can be hundreds or thousands of miles longer than short path 107. Thus, links 113-114 are much more expensive to implement than shorter links 111-112.
In addition to the increased cost, the use of longer links 113-114 adds latency to communications between nodes 105-106. In the above example where router 101 shifts traffic from failed link 111 to link 113, the extra distance of longer link 113 adds latency to communications between nodes 105-106. In additional to the latency added by increased distance, long path 108 typically has more nodes (not shown) in between nodes 105-106 than does short path 107. The higher number of intermediate nodes adds additional latency to communications between nodes 105-106. Many customer applications cannot tolerate the latency of long path 108. The customer may have a Service Level Agreement (SLA) that specifies acceptable latencies.
Thus, current network designs carry large amounts of traffic over long paths—even under normal operating conditions—which forces the network to implement expensive high-capacity links over the longer path. This heavy use of the longer path also adds latency, which forces some customers to use a different communication network.